We appreciate your visit to nature.com. It appears that you are using a browser version that has limited support for CSS. For an optimal browsing experience, we suggest using an updated browser or disabling compatibility mode in Internet Explorer. While we continue to provide the site’s content without styles and JavaScript to ensure continued support, we recommend updating your browser.
The article titled “Seven deadly sins in artificial intelligence for digital medicine” explores the ethical challenges arising as artificial intelligence (AI) becomes more integrated into clinical settings. Despite the swift adoption of AI in medicine, questions regarding trust, fairness, empathy, and governance persist, signifying an unstable ethical landscape. The conceptual framework of the “Seven Deadly Sins of AI in Medicine” identifies recurring systemic failure modes, including blind trust, overregulation, dehumanization, misaligned optimization, overinforming, misapplied statistics, and self-referential evaluation.
Developed through a comprehensive review of scientific literature, clinical guidelines, and regulatory frameworks, the framework aims to address ethical concerns surrounding AI in healthcare. A global opinion poll involving 914 stakeholders from 143 countries between July 2024 and March 2025 validated the framework, showcasing broad agreement on identified risks while highlighting cross-cultural ethical concerns and differing views on regulation between nations.
The article advocates for a shift from fragmented ethical guidelines towards a unified approach by proposing the inversion of the framework into seven cardinal virtues for AI in medicine. These virtues offer actionable principles to guide responsible development and governance of AI in healthcare, emphasizing a trustworthy and human-centered approach. The ultimate goal is to create a unified diagnostic tool to ensure the ethical use of medical AI.
For further details, the article provides access to anonymized data, figures, and analysis scripts via a repository link. The work is supported by grants from the Austrian Science Fund and involves collaborations between various institutions and researchers across different countries. The authors have declared no competing interests.
Published in npj Digital Medicine, the article is available under a Creative Commons Attribution 4.0 International License, allowing sharing, adaptation, distribution, and reproduction with proper credit given to the original authors. To read the full article, access the link provided.
For more insights on AI and robotics research in healthcare, you can sign up for the Nature Briefing newsletter.
About the Author
